At our proposed U54 center, we will continue to conduct epigenomic analysis in cancer. While the focus of the previous award was on epigenetic processes associated with neoplastic transformation of normal cells. In this competing application, we will move a step forward to study epigenetic changes in prostate, breast, and ovarian cancer cells progressing to an aggressive phenotype, i.e., hormone-Zchemo-resistance. Based on our preliminary findings, we hypothesize that epigenetic deregulation of androgen receptor, estrogen receptor a, or TGF-p/SMAD4 signaling underlies the transition of a hormone-Zchemo-sensitive to a hormone- Zchemo-insensitive phenotype in cancer. Different modes of signaling-mediated transcription, including ligand-dependent and -independent functions, will be defined using integrated epigenomic data. We will develop probabilistic algorithms to predict the effect of chromosome looping and chromatin remodeling (i.e., changes of histone marks and DNA methylation) on target gene transcription, including empirical Bayesian mixture and hidden Markov modeling (for classifying spatiotemporal patterns of target genes in a signaling network), interactive modeling of transcription "hubs", stochastic modeling of permissive and non-permissive epigenetic marks, and pattern recognition algorithms for predicting transcription factor binding sites and methylation-prone or -resistant sequences. Testing and validation of these computational predictions will be performed in cancer cell lines. Assays including functional knock-in or -out of key transcription hubs will determine whether cancer cells gain or lose hormone-Zchemo-sensitivity, respectively, as a result of in vitro manipulation. For translational studies, primary tumors will be used to correlate clinicopathological correlations with epigenetic changes. By taking an integrative "omics" approach, we expect to move the epigenomics field forward in at least three new directions: 1) long-range chromatin looping may be a common epigenetic mechanism of transcriptional regulation in cancer;2) histone modificationsZDNA methylation of distant transcription binding sites represent previously uncharacterized biomarkers for predicting hormone-Zchemo-resistance in cancer subtypes;and 3) computational modeling may support the recent notion that repressive histone modifications, rather than DNA methylation, are critical epigenetic factors in the heritable silencing of genes. Importantly, these state-of-the-art computational approaches and the vast omics data will be used for our educationZoutreach efforts to train young systems scientists and for collaborative studies with other researchers in the CCSB-ICBP network.